Microsoft has launched its first cybersecurity-focused AI model alongside a new agentic security platform, marking a direct challenge to rivals Anthropic, Google and OpenAI in the fast-growing field of AI-driven cyber defence. Unveiled at a San Francisco event on Monday, the announcement reflects a broader industry race to build AI tools capable of both attacking and defending software systems, as hackers increasingly weaponise AI in their own campaigns.

The model, called MAI-Cyber-1-Flash, is designed to identify complex vulnerabilities in code and powers Microsoft's existing MDASH vulnerability-detection harness. It underpins a new platform named Perception, which deploys agentic "red", "blue" and "green" teams to simulate attacks, detect bugs and automatically fix them. Microsoft claims the model outperforms rivals including Gemini and various GPT and Mythos versions on the Cyber Gym benchmark, and says tasks once taking hours of specialist work can now be completed in minutes. The tools will enter public preview on 3 November, joining a crowded market that already includes Anthropic's Mythos platform and OpenAI's Day Break security programme.